Exactly once cache framework

• US 7,293,073 B2
• Filed: 07/08/2004
• Issued: 11/06/2007
• Est. Priority Date: 09/06/2001
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: BEA Systems Incorporated (Oracle Corporation)
• Inventors: Jacobs, Dean Bernard, Halpern, Eric M.
• Primary Examiner(s): Ellis; Kevin L.
• Application Number:
  US10/887,690
• Publication Number:
  US 20040243585A1
• Time in Patent Office: 1,216 Days
• Field of Search: 709/213, 709/214, 709/216, 709/217, 711/113
• US Class Current: 709/217
• CPC Class Codes: H04L 67/1095 Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
• CPC Class Codes: H04L 67/288 Distributed intermediate devices, i.e. intermediate devices for interaction with other intermediate devices on the same level
• CPC Class Codes: H04L 67/56 Provisioning of proxy services store-and-forward switching systems in data switching networks H04L12/54
• CPC Class Codes: H04L 67/568 Storing data temporarily at an intermediate stage, e.g. caching
• CPC Class Codes: H04L 69/329 in the application layer [OSI layer 7]
• CPC Class Codes: H04L 69/40 for recovering from a failure of a protocol instance or entity, e.g. service redundancy protocols, protocol state redundancy or protocol service redirection management of faults, events, alarms or notifications in data switching networks H04L41/06
• CPC Class Codes: H04L 9/40 Network security protocols

Abstract

A system for managing objects in a clustered network includes a file system containing at least one copy of a data object. The system can include several clustered servers in communication with the file system. A distributed consensus algorithm is used to select a host server. The selected host server can contain a copy of the data object, such as in local cache, providing access to the local copy to any other server in the cluster. Any change made to an item hosted by the host server can also be updated in the file system. If the host server becomes unable to host the object, a new host can be chosen using the distributed consensus algorithm. The other servers are then notified of the new host by multicast messaging.

Claims

1. A system for managing objects on a network, comprising:

a plurality of network servers, each network server adapted to communicate with a network data source, wherein said plurality of network servers include a distributed consensus algorithm, the distributed consensus algorithm comprising rounds of messages between a lead server and said plurality of network servers, the rounds continuing until a majority of said plurality of network servers agrees on a host server; and

the host server assigned a data object, the data object related to a data item in the network data source, wherein any of said plurality of network servers needing to access the data item can access the data object on the host server.

2. A system according to claim 1, wherein said network servers are selected from the group consisting of hardware cluster servers and software cluster servers.

3. A system according to claim 1, wherein the host server contains a data object comprising a copy of data from the network data source.

4. A system according to claim 1, wherein the host server contains a data object serving as the sole access point for the data item in the network data source.

5. A system according to claim 1, wherein said data item is a transaction log.

6. A system according to claim 1, wherein said distributed consensus algorithm is a Paxos algorithm.

7. A system for managing objects on a network, comprising:

a file system containing at least one copy of a data item;

a plurality of servers in communication with the file system;

a hardware cluster containing hardware cluster servers located in said plurality of servers, said hardware cluster containing a distributed consensus algorithm for selecting a lead server from among said hardware cluster servers; and

a host server in said plurality of servers, said host server containing a local copy of the data item, said host server adapted to provide access to the local copy to any of said plurality of servers and update the copy of the data item in the file system any time an update is made to the local copy.

8. A system according to claim 7, wherein said host server is in said hardware cluster.

9. A method for assigning ownership of an object on a network, comprising:

selecting a host server from among a plurality of network servers using a distributed consensus algorithm stored on a hardware cluster;

assigning a data object to the host server, the host server assigned to provide sole access to a data object on the network;

pulling a copy of a data object from a file system to the host server; and

notifying other network servers that the host server contains a copy of the data object to be used in processing network requests.